A crowdsourced map of the U.S. AI policy landscape, tracking the people, organizations, and resources shaping AI governance.
Live site: mapping-ai.org GitHub: MappingAI/mapping-ai
Mapping AI is a collaboratively maintained database of actors in U.S. AI policy: legislators, regulators, researchers, funders, advocates, frontier labs, and civil society organizations. The goal is to identify who is shaping AI governance, where coalitions are forming, and where the gaps are, then use that map as the foundation for a coordinated progressive policy agenda.
The project is maintained by a working group of researchers, policy experts, and practitioners. Public submissions are welcome.
Frontend: Vite 8 MPA + React 19 + TypeScript + Tailwind CSS v4. The stakeholder map (map.html) is inline D3.js + Canvas 2D, not React.
Backend and infrastructure specifics are documented in docs/architecture/current.md. A migration off AWS to Cloudflare Workers + Neon + TanStack Start is in progress; see docs/architecture/target.md and ADR-0001.
Prerequisites: Node.js 20+ and pnpm. Install Node via nvm or brew install node, then brew install pnpm (or npm install -g pnpm).
git clone https://github.com/MappingAI/mapping-ai.git
cd mapping-ai
pnpm install --frozen-lockfile
brew install lefthook
lefthook installCreate a .env file in the project root with database credentials (shared via Doppler, or ask the team):
# .env
DATABASE_URL=postgresql://...
AWS_ACCESS_KEY_ID=...
AWS_SECRET_ACCESS_KEY=...pnpm run devVisit localhost:5173. This runs Vite and the Express API server together. Everything works from there: map, contribute form, admin, insights, search, form submissions.
On first run, map-data.json is generated from your database automatically (needs DATABASE_URL in .env). If you don't have DB credentials, download the production snapshot first so the map isn't empty:
curl -o map-data.json https://mapping-ai.org/map-data.json
curl -o map-detail.json https://mapping-ai.org/map-detail.jsonIf you're using Claude Code, Cursor, or similar:
- Read
CLAUDE.mdfirst for full codebase context - Use
npx tsc --noEmitto catch type errors after changes - Use
agent-browserfor visual testing (screenshots, form interaction) npx vitest runto verify tests- The compound-engineering plugin has useful skills:
/ce:reviewfor code review,/ce:compoundfor documentation
Visit mapping-ai.org/contribute to submit a person, organization, or resource. All submissions are reviewed before appearing on the map.
What we track:
- People: policymakers, researchers, funders, advocates, journalists actively shaping AI policy
- Organizations: frontier labs, think tanks, government agencies, labor groups, civil society, academic institutions
- Resources: essays, reports, books, podcasts, and academic papers relevant to AI governance
Submission fields include regulatory stance, influence type, threat model beliefs, AGI timeline views, and source evidence, so the map captures not just who is involved but where they stand.
- docs/architecture/current.md - architecture, API reference, schema, deployment
- docs/architecture/target.md - planned architecture (migration in progress, see ADR-0001)
- ONBOARDING.md - setup guide for new contributors
- CLAUDE.md - codebase conventions and AI assistant context
- docs/DEPLOYMENT.md - deploy process and review guidelines
Join the Discord server for discussions, coordination, and mapping parties. Use the forum channels to track bugs, share research questions, and post feature ideas.
Maintained by a working group of researchers, policy experts, and practitioners. See mapping-ai.org/about for more.
This project is maintained by the Mapping AI Working Group. Content is (c) 2026 the respective contributors.